Re: r38000 group

On Wed, 2009-09-23, Julian Foad wrote:
> I still have some concerns about this back-port branch.
>
> I have just gone through the branch and made a diff that represents the
> whole state of the branch, and I have started to go through it and in so
> doing I have written a log message based purely on what code changes I
> see. I want to compare this with the original log messages of the
> individual revisions, and see what light that throws on it, but I
> haven't done so yet and it's end of day for me now.

Another thing: r39053 makes a merge test XFail due to "duplicate
notifications" with RA Serf. r39156 fixes "a duplicate notification" but
says "(Test still XFails for various status imperfections.)".

Should the test file at least be updated to no longer say...

    # ra_serf causes duplicate notifications with this test:
    Skip(merge_replace_causes_tree_conflict,
         svntest.main.is_ra_type_dav_serf),

?

I know it's "just" a comment. Really I was concerned about why we care
about differences between RA Serf and other RA layers. Are they bugs or
differences that we should tolerate?

Here are the log messages, with the original log messages inserted:

------------------------------------------------------------------------
r39053 | stsp | 2009-09-02 00:23:59 +0100 (Wed, 02 Sep 2009) | 2 lines

merge r39052 from trunk.

r39052 | stsp | 2009-09-02 00:14:57 +0100 (Wed, 02 Sep 2009) | 4 lines

* subversion/tests/cmdline/merge_tests.py
  (test_list): Skip merge_replace_causes_tree_conflict if using ra_serf.
    This test does not pass with ra_serf because of a duplicate notification.

------------------------------------------------------------------------
r39165 | sbutler | 2009-09-07 16:55:16 +0100 (Mon, 07 Sep 2009) | 5 lines

merge -c 39096,39124,39125,39127 from trunk.

Fix a duplicate notification for merge with incoming file replace.
Rearrange merge notification code for clarity.

r39096 | neels | 2009-09-03 02:49:31 +0100 (Thu, 03 Sep 2009) | 7 lines

* subversion/libsvn_client/repos_diff.c
  (close_file): Fix duplicate notification for tree-conflict with incoming
    replace (file replaced with file), possibly others. This case is visible
    in merge_tests.py merge_replace_causes_tree_conflict2 (135) on file
    A/mu during the final merge: there was an additional "A" notification
    for A/mu. (Test still XFails for various status imperfections.)

r39124 | neels | 2009-09-03 23:25:01 +0100 (Thu, 03 Sep 2009) | 5 lines

* subversion/libsvn_client/repos_diff.c
  (close_file): Rearrange notification code. Don't bother setting NOTIFY's
    LOCK_STATE in case of a tree-conflict notification.

r39125 | neels | 2009-09-03 23:35:57 +0100 (Thu, 03 Sep 2009) | 1 line

* subversion/libsvn_client/repos_diff.c: Comment.

r39127 | neels | 2009-09-04 00:29:30 +0100 (Fri, 04 Sep 2009) | 5 lines

* subversion/libsvn_client/repos_diff.c
  (add_directory): Rearrange notification code (no functional change, but
    it looks safer this way).
  (close_file): Comment.

------------------------------------------------------------------------

- Julian

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2399112
Received on 2009-09-24 02:32:24 CEST